Abstract

The Vocational Training Facility (VTF) has developed a three-workstation classroom as part of a project to test a new approach to the vocational training of students with physical disabilities. The three workstations are equipped with an array of devices and software to facilitate the access of curriculum and daily-living materials to the students. One of the devices in each workstation is the Desktop Vocational Assistant Robot (DeVAR), developed over a period of four years prior to VTF, and tested in clinical and vocational settings by users with high-level quadriplegia. This article describes the use of DeVAR in a classroom setting, and how the tasks and settings compare to previous environments.

Background

The Vocational Training Facility

The Vocational Training Facility (VTF) has developed and is evaluating a learning environment for students with physical disabilities. The goal is to eliminate the barriers of access to the tools and materials involved in learning and working [Hammel, et al., in press]. The curriculum, a 3-month intensive course on Desktop Publishing (DTP), is therefore entirely computer-based and controlled. Learning materials, programmed on an Apple Macintosh(TM), include a HyperCard(TM) stack for student training and performance assessment, computer-aided instruction (CAI) and laserdisc information controlled by the student and presented in real-time on the computer screen. Each student receives access software (e.g., Easy Access(TM), phone handling program, macros and utilities) and low to high technology access devices (e.g., mouthstick, trackball, adjustable keyboard mount, ultrasonic head-position control for cursor motion, voice recognition unit, tongue-touch keypad, voice-controlled robot). These ensure access to the learning materials and to items required for activities of daily living (ADLs) during the training sessions.

Student selection

The VTF is evaluating the effectiveness of the curriculum for students with high level quadriplegia (C2-C5). Students with low-level quadriplegia (C6-C8), who have limited arm function, and with paraplegia, who have complete hand and arm function, form the other two test subject groups. In total, the VTF will have trained six people from each of the three groups over a 24-month period. The robot is primarily intended to be used by the first group. This paper discusses the DeVAR-related results from the first four students with high-level quadriplegia, and its limited use by two students with low-level quadriplegia.

DeVAR tasks for VTF

Each of the three VTF workstations includes a Desktop Vocational Assistant Robot (DeVAR), which is composed of a Stäubli PUMA-260(TM) robot arm mounted on a 1 m. long overhead transverse track [Van der Loos, et al., 1989]. DeVAR has a servo-controlled gripper, an Otto-Bock Greifer, capable of grasping a 10 cm. wide, 1 kg. object with a 15 kg. grip. It can handle most common objects found on the desktop.

The vocational task set for VTF includes: Macintosh 3.5" diskette handling; laserdisc handling; paper handling from printer, to/from copystands, to/from holding slots and to an outbox; phone line dialing and answering via speakerphone or handset; and mouthstick handling. ADLs include: drink cup handling (or manipulating a flex straw from a pitcher); lunch preparation and presentation of a microwavable or cold meal; feeding using a spoon or fork; medication and throat lozenge selection and presentation using spoons; and nose scratching with a plastic fork. Each task has interactive phases to allow the user to select options and objects, and automatic phases for arm, track and gripper motion.

DeVAR is controlled through a discrete-word, speaker-dependent voice recognition unit, a Votan VPC-2100. Each user takes about 15 minutes to train a template file of the 60 vocabulary words, and typically trains a new file several times during the first week of learning to use the robot. Subsequently, single words are retrained on an as-needed basis.

Objective

Providing skills for empowerment

The most important qualitative objective of VTF is that each student becomes aware of opportunities previously out of reach, and acquires tools to begin exploration of personal vocational goals. Central to this VTF objective is the introduction to a vocational field (DTP) that has potential for personal growth, both economically and professionally. DTP is also a field for which data entry per se is not a determinant of overall job performance, an important aspect for competitiveness with other job applicants. Beyond this, the VTF teaches job skills and communication skills that supplement those learned during rehabilitation following the initial injury.

These factors are aimed at providing the knowledge to compete effectively in the job market, bolstering the incentive to rejoin the workforce, and strengthening the resolve to emerge from disability income support systems and become self-supportive.

Providing exposure to technology

A person with a disability reentering the workforce is often faced with the need to educate supervisors and colleagues on the job-specific functional aspects of the disability. Beyond communication skills, an important component of VTF is exposure to available methods, devices, adaptations, and computer-technology tools. The knowledge to help the employer in satisfying the "reasonable accommodation" requirement can be a great asset, and allows the implementation of a comfortable, functional work environment at the lowest cost and with the least waste of energy, time and resources.

The VTF exposes each student to the array of appropriate market-available tools and a smaller set of prototype devices and systems, such as DeVAR. Each student can assess, based on actual use, which devices are preferable and functional on a long-term basis. The VTF staff can also assess the value of the prototype devices and how they might be modified for greater functionality in the future.

Approach

Workstation set-up

The VTF has an established procedure for admitting students [Hammel & Van der Loos, 1991]. Once admitted, each student is interviewed by an occupational therapist and rehabilitation engineer prior to the start of the 12-week training session. The level of the student's function determines if DeVAR will be used. If not, the workstation components are arranged in accordance with the student's range of motion, wheelchair specifications, level of functional independence in task performance and personal preferences.

Robot programming

If DeVAR is used, the workstation itself is not modified substantially, since it is already arranged in accordance with the robot's range of motion and geometry. The set-up of DeVAR for VTF is not substantially different from previous configurations [Hammel, et al., 1992]. In particular, the user interface remains constant. However, certain aspects involving wheelchair height, task preference, and particular needs are discussed beforehand. Certain robot tasks involve bringing objects to the user, such as spoons, mouthstick and phone receiver. Each task is adjusted by the staff so that these items are brought to the correct positions. If additional tasks are needed, they are programmed, tested, and incorporated into the user interface.

Results

VTF users and environment

DeVAR was designed to replace an attendant, whose role was to perform user and robot set-up two to three times a day, and who was absent otherwise. In the VTF classroom, the continuous presence of a teacher means a de facto assistant at all times as a robot back-up. In addition, with its three workstations, the classroom is a more interactive setting than a single-user office environment. These differences in environment have resulted in differences in DeVAR usage. Overall, there is less reliance on DeVAR.

Differences between anticipated and actual usage

During the first year of VTF training sessions, DeVAR has been used by the students with high and low level quadriplegia in different ways than originally anticipated:

* Two of the students with low level quadriplegia had residual arm function, and used DeVAR for some tasks (i.e., medication and phone), and other, low-technology accommodations for others (e.g., paper handling, laserdiscs).

* Two students with high level quadriplegia have had a full-time attendant, and have used the robot and attendant interchangeably.

* Since the workstations are in a classroom and not an enclosed office, the ambient noise at times interferes with voice recognition in a manner not anticipated in the original design, causing periodic frustration. Experimentation with microphone placement was needed to reduce the influence of noise.

* To provide phone line access, the classroom has one shared line. Each DeVAR has a hands-off, built-in phone line handling feature, which works adequately for a single-line setting. However, it was not intended to be used cooperatively, causing problems when calls have to be transferred to one of the other workstations.

* None of the students wished to use the robot's lunch serving tasks on a routine basis. A total of only two meals have been served to date. Partly this was due to the limited menu (only liquid or canned servings such as soup, applesauce and chili), but mostly because lunch is a social event, a time of the day to get out of the classroom, take a break, and talk to people. Eating alone in the classroom was not acceptable.

* The medication and throat lozenge tasks were well-received, especially because the students could request these ADL tasks at any time during the day, not just when someone happened to be near to help. However, the robot's drink task took more than 60 seconds to bring the cup to the user's mouth. This was deemed too slow a response due to the unpleasant taste of the medication. Therefore, most students used an always-accessible water pitcher and long straw.

Task changes requested by students

Most students with high or low level quadriplegia and adequate neck range of motion preferred a water pitcher and long rigid straw over a cup brought out of the cooler by the robot, especially when using medication. One user with limited neck motion asked for a 60 cm. long rubber tube, used as a flexible straw, to be inserted in the water pitcher. The robot then brought the other end of the tube, held in a clip, to his mouth. This was only marginally successful due to the high suction required.

Two other modifications involved telephone use. One user brought his own dialing and answering device, provided by the Pacific Bell Telephone Company, which was set up by his attendant in the morning. Another user wanted to set up a rolodex-type file on the Macintosh. The shareware program AddressBook(TM) was tried out temporarily, in conjunction with a modem, to dial out. This accommodation proved successful.

Discussion

Reasons for usage preference

None of the targeted DeVAR users elected to use the robot for all available tasks. Contributing factors were the classroom setting, continuous availability of assistance by the instructor and attendants, difficulties with the robot itself (e.g., voice recognition), and learning the DeVAR interface. The overriding reason for a lack of intensive DeVAR use, however, is that students are enrolled in the VTF primarily to learn DTP, and secondarily to acquire knowledge about market-available assistive devices. It is only a third project priority to expose students to prototype devices. It is of more immediate relevance to teach students about effective acquisition and use of attendant care in the workplace.

Future directions

The use of DeVAR in this new multi-workstation setting has shown that several modifications would make the robot easier to use. These include:

* integrating the interface with the primary computer, the Macintosh, so that reliance on voice recognition is no longer critical, and so that the interface adopts the Macintosh "look-and-feel";

* rethinking the phone interface, so that it is no longer a robot-mediated, but rather a Macintosh-controlled feature;

* adding sensors to the gripper to better accommodate objects slightly out of place.

It is not within the scope of the VTF to make these changes. However, the ongoing VA clinical evaluation of DeVAR, coordinated nationally by the VA Technology Transfer Section, will make its own recommendations in 1994 [Interim report: Cupo, 1993]. Future DeVAR development would therefore have two sources of extensive user feedback to consult in an effort to derive a market-ready, effective robotic assistant for students and office workers with severe physical disabilities.

References

M. Cupo, Clinical Evaluation of the Desktop Vocational Assistant Robot DeVAR. Interim Report, VA Technology Transfer Section, Baltimore, MD, November, 1993.

J. Hammel, H.F.M. Van der Loos, Factors in the Prescription and Cost-Effectiveness of Robot Systems for High-Level Quadriplegics. Proc. 14th Annual RESNA Conference, Kansas City, MO, June, 1991.

J. Hammel, H.F.M. Van der Loos, I. Perkash, Evaluation of a Vocational Robot with a Quadriplegic Employee. Archives of Physical Medicine and Rehabilitation, Vol 73, July, 1992, pp. 683-693.

J. Hammel, P. LePage, I. Perkash, C. Burgar, D. Shafer, E. Topp, D. Lees, H.F.M. Van der Loos, The Vocational Training Facility: An Interactive Learning Program to Return Persons with Physical Disabilities to Employment. Work: A Journal of Prevention, Assessment and Rehabilitation. (in press)

H.F.M. Van der Loos, J. Hammel, D. Schwandt, D. Lees, L. Leifer, I. Perkash, Design and Evaluation of a Vocational Desktop Robot. Proc. 12th Annual RESNA Conference, New Orleans, June, 1989.

Acknowledgments

Support for this project is provided by the U.S. Dept. of Veterans Affairs Merit Review Grant B-635 and by Core funding of the Rehabilitation R&D Center of the Palo Alto VA Medical Center. We wish to thank the VTF students and instructors for their contributions to the project.

Author's Address

H.F. Machiel Van der Loos, Ph.D.
Rehabilitation R&D Center
VA Medical Center
3801 Miranda Ave., MS 153
Palo Alto, CA  94304-1200

Citation (Copyright RESNA Press, 1994)

H.F.M. Van der Loos, J. Hammel, Use of a rehabilitation robot in a classroom setting.